Docling is an open-source, MIT-licensed toolkit that converts PDFs and many other document formats into clean Markdown or JSON, keeping layout, reading order and tables.
In plain words
Docling uses specialised AI models to understand page layout and table structure, then outputs a structured document you can export as Markdown, HTML or JSON. It also reads Word, PowerPoint, Excel, HTML, images, audio and more. It runs on ordinary hardware and connects to generative AI tools, and it is hosted by the LF AI & Data Foundation.
A quick example
The Docling README shows one command that takes the web address of an arXiv PDF and writes a Markdown file of its content in the current folder. That Markdown, with headings and tables intact, can then be split into chunks for a RAG system.
Where you'll meet it
- Preparing PDFs and office files for RAG.
- Extracting tables from reports and papers.
Related
Sources
- Docsdocling-project/docling, Docling project (GitHub) · read 4 Oct 2026
- PaperDocling Technical Report, Auer et al. (arXiv) · read 29 Sept 2026
- DocsDocling documentation, Docling project · read 29 Sept 2026
Checked by WhatIsAI against the sources above · 4 Oct 2026. Spotted an error? Report it.